1. Field of the Invention
The present invention relates to the delivery of DNA to and the expression of delivered genes in, cells of the nervous system.
2. Description of the Related Art
The first human gene therapy trial started in September 1990 and involved retrovirally-mediated transfer of the adenosine deaminase (ADA) gene into lymphocytes of patients with severe combined immunodeficiency (SCID). The, favorable results of this trial stimulated further interest in gene therapy resulting in further 67 gene therapy clinical protocols approved by the NIH Recombinant DNA Advisory Committee (RAC) to date. Although the original promise of gene therapy was the development of a curative treatment for simple, single gene diseases, the vast majority of gene therapy trials have been for complex genetic or acquired diseases, such as infectious disease and cancer. A large number of the initial clinical gene transfer studies were not gene therapy but rather gene marking studies. The first type of marking experiments used tumor infiltrating lymphocytes which were transduced in vitro with retroviral vectors prior to infusion into patients with cancer. The second class of gene marking studies involved the attempt to detect residual tumor cells in marrow infused into patients following ablative chemotherapy.
Of the currently approved gene therapy trials, all trials prior to 1992 used retroviral vectors and the diseases targeted included SCID, familial hypercholesteremia and cancer. More recently, gene therapy trials have commenced for AIDS and Hemophilia B, again using retroviral vectors. In addition, adenoviral vectors have recently been approved for cystic fibrosis. The vast majority of these protocols have enrolled very few patients at present and most of the trials are as yet unpublished. The available data however appears promising, for example the expression of the LDL receptor in the liver following ex vivo transduction of resected hepatocytes and its infusion into the portal vein in patients with familial hypercholesteremia has resulted in a 20% drop in plasma cholesterol levels (Randall (1992) JAMA 269:837-838). It is likely therefore that there will be an exponential growth in gene therapy trials and a large number of medical schools and teaching hospitals are setting up gene therapy centers.
The ability to deliver genes to the nervous system, and to manipulate their expression, may make possible the treatment of numerous neurological disorders. Unfortunately, gene transfer into the central nervous system (CNS) presents several problems including the relative inaccessibility of the brain and the blood-brain-barrier, and that neurons of the postnatal brain are post-mitotic. The standard approach for somatic cell gene transfer, i.e., that of retroviral vectors, is not feasible for the brain, as retrovirally mediated gene transfer requires at least one cell division for integration and expression. A number of new vectors and non-viral methods have therefore been used for gene transfer in the CNS. Although the first studies of gene transfer in the CNS used an ex vivo approach, i.e., the transplantation of retrovirally-transduced cells, more recently several groups have also used an in vivo approach. Investigators have used HSV-1 and adenoviral vectors as well as non-viral methods including cationic lipid mediated transfection (Wolff (1993) Curr. Opin. Biol. 3:743-748).
The ex vivo approach is illustrated by a recent study in which oligodendrocytes were retrovirally infected and transplanted into a syngeneic rat model for demyelination (Groves et al (1993) Nature 362:453-457). In addition to the use of brain cells as vehicles for foreign gene expression in the CNS, non-neuronal cells including fibroblasts and primary muscle cells have also been used (Horrelou et al (1990) Neuron 5:393-402; Jiao et al (1993) Nature 362:450-453).
The in vivo approach was initially largely based on the use of the neurotropic Herpes Simplex Virus (HSV-1), however, HSV vectors present several problems, including instability of expression and reversion to wild-type (see below). A more recent development has been the use of adenoviral vectors. Adenoviral vector studies have shown expression of marker genes into the rat brain persisting for two months although expression fell off dramatically (Davidson et al (1993) Nature Genetics 3:219-2223). In addition to viral vector approaches, other investigators have used direct injection of a cationic liposome:plasmid complex obtaining low level and transient expression of a marker gene (Ono et al (1990) Neurosci. Lett. 117:259-263).
There have been very few studies using xe2x80x9ctherapeuticxe2x80x9d genes in the CNS. The majority of these have used the ex vivo approach with transduction of fibroblasts and muscle cells with the human tyrosine hydroxylase gene in order to produce L-dopa-secreting cells for use in models of Parkinson""s Disease (e.g., Horrelou et al (1990) Neuron 5:393-402; Jiao et al (1993) Nature 362:450-453). Of the in vivo approaches, HSV vectors have been used to express xcex2-glucuronidase (Wolfe et al (1992) Nature Genetics 1:379-384), glucose transporter (Ho et al (1993) Proc. Natl. Acad. Sci. 90:6791-6795) and nerve growth factor (Federoff et al (1992) Proc. Natl. Acad. Sci. 89:1636-1640). An adenoviral vector has been used to induce low level transient expression of human xcex11-antitrypsin (Bajoccchi et al (1993) 3:229-234).
The only clinical studies of gene transfer in the brain followed a report by Culver et al (1992) Science 256:18550-18522) in which they essentially cured rats which had been intracerebrally implanted with glioma cell lines. They used a retrovirus expressing the HSV-1 thymidine kinase (tk) gene and then subsequently treated with ganciclovir. In 1993, a human protocol for glioblastoma multiforme was approved using the retroviral tk vectorxe2x80x94ganciclovir protocol (Oldfield et al (1993) Human Gene Ther. 4:39-69).
Herpes Viruses The genome of the herpes simplex virus type-1 (HSV-1) is about 150 kb of linear, double-stranded DNA, featuring about 70 genes. Many viral genes may be deleted without the virus losing its ability to propagate. The xe2x80x9cimmediately earlyxe2x80x9d (IE) genes are transcribed first. They encode trans-acting factors which regulate expression of other viral genes. The xe2x80x9cearlyxe2x80x9d (E) gene products participate in replication of viral DNA. The late genes encode the structural components of the virion as well as proteins which turns on transcription of the IE and E genes or disrupt host cell protein translation.
After viral entry into the nucleus of a neuron, the viral DNA can enter a state of latency, existing as circular episomal elements in the nucleus. While in the latent state, its transcriptional activity is reduced. If the virus does not enter latency, or if it is reactivated, the virus produces numerous infectious particles, which leads rapidly to the death of the neuron. HSV-1 is efficiently transported between synaptically connected neurons, and hence can spread rapidly through the nervous system.
Two types of HSV vectors have been utilized for gene transfer into the nervous system. Recombinant HSV vectors involve the removal of an immediate-early gene within the HSV genome (ICP4, for example), and replacement with the gene of interest. Although removal of this gene prevents replication and spread of the virus within cells which do not complement for the missing HSV protein, all of the other genes within the HSV genome are retained. Replication and spread of such viruses in vivo is thereby limited, but expression of viral genes within infected cells continues. Several of the viral expression products may be directly toxic to the recipient cell, and expression of viral genes within cells expressing MHC antigens can induce harmful immune reactions. In addition, nearly all adults harbor latent herpes simplex viruses within neurons, and the presence of recombinant HSV vectors could result in recombinations which can produce an actively replicating wild-type virus. Alternatively, expression of viral genes from the recombinant vector within a cell harboring a latent virus might promote reactivation of the virus. Finally, long-term expression from the recombinant HSV vector has not been reliably demonstrated. It is likely that, except for conditions in which latency is induced, the inability of HSV genomes to integrate within host DNA results in susceptibility to degradation of the vector DNA.
In an attempt to circumvent the difficulties inherent in the recombinant HSV vector, defective HSV vectors were employed as gene transfer vehicles within the nervous system. The defective HSV vector is a plasmid-based system, whereby a plasmid vector (termed an amplicon) is generated which contains the gene of interest and two cis-acting HSV recognition signals. These are the origin of DNA replication and the cleavage packaging signal. These sequences encode no HSV gene products. In the presence of HSV proteins provided by a helper virus, the amplicon is replicated and packaged into an HSV coat. This vector therefore expresses no viral gene products within the recipient cell, and recombination with or reactivation of latent viruses by the vector is limited due to the minimal amount of HSV DNA sequence present within the defective HSV vector genome. The major limitation of this system, however, is the inability to eliminate residual helper virus from the defective vector stock. The helper virus is often a mutant HSV which, like the recombinant vectors, can only replicate under permissive conditions in tissue culture. The continued presence of mutant helper HSV within the defective vector stock, however, presents problems which are similar to those enumerated above in regard to the recombinant HSV vector. This would therefore serve to limit the usefulness of the defective HSV vector for human applications.
For further information on HSV-mediated gene delivery to neurons, see Breakefield and DeLuca, xe2x80x9cHerpes Simplex Virus for Gene Delivery to Neurons,xe2x80x9d (1991) New Biologist 3:203-18; Ho and Mocarski (1988) xe2x80x9cBeta-Galactosidase as a marker in the herpes simplex virus-infected mouse,xe2x80x9d Virology 167:279-93; Palella, et al (1988) xe2x80x9cHerpes Simplex Virus-Mediated human hypoxanthine-guanine phosphoribosyl-transferase gene transfer into neuronal cells,xe2x80x9d Molec. and Cell. Biol. 8:457-60; Pallela et al (1988) xe2x80x9cExpression of human HPRT mRNA in brains of mice infected with a recombinant herpes simplex virus-1 vector,xe2x80x9d Gene 80:137-144; Andersen et al (1992) xe2x80x9cGene transfer into mammalian central nervous system using the neuron-specific enolase promoter,xe2x80x9d Human Gene Therapy 3:487-99; Kaplitt et al (1993) xe2x80x9cMolecular alterations in nerve cells: Direct manipulation and physiological mediation,xe2x80x9d Curr. Topics Neuroendocrinol. 11:169-191; Spaele and Frenkel (1982) xe2x80x9cThe Herpes Simplex Virus Amplicon: A New Eukaryotic Defective-Virus-Cloning-Amplifying Vector,xe2x80x9d Cell 30:295-304 (1982); Kaplitt et al (1991) xe2x80x9cExpression of a Functional Foreign Gene in Adult Mammalian Brain Following In Vivo Tranfers via a Herpes-Simplex Virus. Type 1 Detective Viral Vector,xe2x80x9d Molec. and Cell. Neurosci. 2:320-30; Federoff et al (1992) xe2x80x9cExpression of Nerve Growth Factor In Vivo form a Defective Herpes Simplex Virus 1 Vector Prevents Effects of Axotomy on Sympathetic Ganglia,xe2x80x9d Proc. Nat. Acad. Sci. (USA) 89:1636-40.
While HSV vectors of reduced toxicity and replication ability have been suggested, they can still mutate to a more dangerous form, or activate a latent virus, and, since the HSV does not integrate, achieving long-term expression would be difficult.
Adenoviruses
The adenovirus genome consists of about 36 kb of double-stranded DNA. Adenoviruses target airway epithelial cells, but are capable of infecting neural cells.
Recombinant adenovirus vectors have been used as gene transfer vehicles for non-dividing cells. These vectors are similar to recombinant HSV vectors, since the adenovirus E1a immediate-early gene is removed but most viral genes are retained. Since the E1a gene is small (roughly 1.5 kb) and the adenovirus genome is ⅓ the size of the HSV genome, other non-essential adenovirus are removed in order to insert a foreign gene within the adenovirus genome.
In nature, diseases resulting from adenovirus infections are not as severe as those induced by HSV infection, and this is the principal advantage of recombinant adenovirus vectors compared with HSV vectors. However, retention and expression of many adenovirus genes presents problems similar to those described with the HSV vector, particularly the problem of cytotoxicity to the recipient cell. In addition, recombinant adenovirus vectors often elicit immune responses which may serve to both limit the effectiveness of vector-mediated gene transfer and may provide another means for destruction of transduced cells. Finally, as with the HSV vectors, stability of long-term expression is currently unclear since there is no mechanism for specific viral integration in the genome of non-dividing host cells at high frequency. While theoretically possible, defective adenovirus vectors would be difficult to make as at least 20% of the Ad genome is required for packaging (about 27 kb) and vectors this size are difficult to work with. In contrast, the defective HSV vectors are small plasmids which replicate until the correct aggregate size is reached for proper packaging.
For more information on vectors, see Akli et al (1993) xe2x80x9cTransfer of a foreign gene into the brain using adenovirus vectors,xe2x80x9d Nature Genetics 3:224-228; La Salle, et al, xe2x80x9cAn adenovirus vector for gene transfer into neurons and glia in the brain,xe2x80x9d Science 259:988-90 (1993), Editorial, xe2x80x9cAdventures with adenovirus,xe2x80x9d 3:1-2 (1993); Neve, xe2x80x9cAdenovirus vectors enter the brainxe2x80x9d TIBS 16:251-253 (1993).
Adeno-Associated Virus is a defective parvovirus whose genome is encapsidated as a single-stranded DNA molecule. Strands of plus and minus polarity are both packaged, but in separate virus particles. Although AAV can replicate under special circumstances in the absence of a helper virus, efficient replication requires coinfection with a helper virus of the herpesvirus or adenovirus family. In the absence of the helper virus, AAV establishes a latent infection in which the viral genome exists as an integrated provirus in the host cell. (No AAV gene expression is required to establish a latent infection). The integration of the virus is site-specific (chromosome 19). If a latently infected cell line is later superinfected with a suitable helper virus, the AAV provirus is excised and the virus enters the xe2x80x9cproductivexe2x80x9d phase of its life cycle. However, it has been reported that certain AAV-derived transducing vectors are not rescued by adenovirus superinfection.
Although AAV is a human virus, its host range for lytic growth is unusually broad. Cell lines from virtually every mammalian species tested (including a variety of human, simian, canine, bovine and rodent cell lines) can be productively infected with AAV, provided an appropriate helper virus is used (e.g., canine adenovirus in canine cells). Despite this, no disease has been associated with AAV in either human or other animal populations, unlike both HSV and adenovirus.
AAV has been isolated as a nonpathogenic coinfecting agent from fecal, ocular and respiratory specimens during acute adenovirus infections, but not during other illnesses.
Likewise, latent AAV infections have been identified in both human and nonhuman cells. Overall, virus integration appears to have no apparent effect on cell growth or morphology. See Samulski (1993) Curr. Op. Gen. Devel. 3:74-80.
The genome of AAV-2 is 4,675 bases in length and is flanked by inverted terminal repeat sequences of 145 bases each. These repeats are believed to act as origins for DNA replication.
There are two major open reading frames. The left frame encodes at least four non-structural proteins (the Rep group). There are two promoters P5 and P19, which control expression of these proteins. As a result of differential splicing, the P5 promoter directs production of proteins Rep 78 and Rep 68, and the P19 promoter, Rep 52 and Rep 40. The Rep proteins are believed to be involved in viral DNA replication, trans-activation of transcription from the viral promoters, and repression of heterologous enhancers and promoters.
The right ORF, controlled by the P40 promoter, encodes the capsid proteins Vp1 (91 kDa), Vp2 (72 kDa) and Vp3 (60 kDa). Vp3 comprises 80% of the virion structure, while Vp1 and Vp2 are minor components. There is a polyadenylation site at map unit 95. For the complete sequence of the AAV-2 genome, see Vastava et al (1983) J. Virol. 45:555-64.
McLaughlin et al ((1988) J. Virol. 62:1963-73) prepared two AAV vectors: dl 52-91, which retains the AAV rep genes, and dl 3-94, in which all of the AAV coding sequences have been deleted. It does, however, retain the two 145 base terminal repeats, and an additional 139 bases which contain the AAV polyadenylation signal. Restriction sites were introduced on either side of the signal.
A foreign gene, encoding neomycin resistance, was inserted into both vectors. Viral stocks were prepared by complementation with a recombinant AAV genome, which supplied the missing AAV gene products in trans but was itself too large to be packaged.
Unfortunately, the virus stocks were contaminated with wild type AAV (10% in the case of dl 3-94) presumably as a result of homologous recombination between the defective and the complementing virus.
Samulski et al ((1989) J. Virol. 63:3822-28) developed a method of producing recombinant AAV stocks without detectable wild-type helper AAV. Their AAV vector retained only the terminal 191 bases of the AAV chromosome. In the helper AAV, the terminal 191 bases of the AAV chromosome were replaced with adenovirus terminal sequences. Since sequence homology between the vector and the helper AAV was thus essentially eliminated, no detectable wild-type AAV was generated by homologous recombination. Moreover, the helper DNA itself was not replicated and encapsidated because the AAV termini are required for this process. Thus, in the AAV system, unlike the HSV system, helper virus could be completely eliminated leaving a helper-free AAV vector stock.
Muro-Cacho et al ((1992) J. Immunother. 11:231-237) have used AAV-based vectors for gene transfer into both T- and B-lymphocytes. Walsh et al ((1992) Proc. Nat. Acad. Sci. (USA) 89:7257-61) used an AAV vector to introduce a human gamma globulin gene into human erythroleukemia cells; the gene was expressed. Flothe et al ((1993) J. Biol. Chem. 268:3781-90) delivered the cystic fibrosis transmembrane conductance regulator gene to airway epithelial cells by means of an AAV vector. See also Flotte et al (1992) Am. J. Respir. Cell. Mol. Biol. 7:349-56; Flotte et al (1993) Proc. Nat. Acad. Sci. (USA) 90:10613-17.
Adeno-associated virus has not been reported to naturally infect any nervous system cells, and AAV-derived vectors have not previously been used to transfect terminally differentiated, non-dividing cells. Nonetheless, the present invention demonstrates that an adeno-associated virus-derived vector may be used to deliver exogenous DNA to cells of the postnatal central and/or peripheral nervous system, including neurons and glia, even though these cells are non-dividing. Specificity may be achieved by anatomically specific delivery or by tissue specific expression.
The exogenous DNA preferably comprises a gene which encodes a gene product useful in the treatment of a nervous system disorder. This gene, in some embodiments, is operably linked to a promoter specific for particular cell types or regions within the nervous system. Because the AAV vector is integrated, stable, longterm expression (e.g., for greater than seven months) can be achieved.